Physical properties of linear-chain systems. III. Absorption spectra of

T E Wood , A Muirhead , P Day. Journal of Physics C: Solid State Physics 1978 11 (8), 1619-1633. E Krausz , S Viney , P Day. Journal of Physics C: Sol...
0 downloads 0 Views 812KB Size
826 Inorganic Chemistry, Vol. 15, No. 4, 1976

Finally, considering possible interchain interactions, one notes that the closest approach of the two different types of chains is about the same as the closest approach between the (MnC14)2- tetrahedral chains. It thus seems likely that interchain interactions between the two different types of chains will be responsible for any long-range transition which occurs in [(CH3)3NH]3MnzC17. Because of the small intrachain exchange in the tetrahedral chains, it would seem that there is a reasonable possibility that such a transition will occur at a lower temperature than the one observed in TMMC. In any case, lower temperature measurements should provide useful insight into the interchain interactions and may well prove [(CH3)3NH]3Mn2C17 to be a more ideal one-dimensional material than TMMC.

Acknowledgment. The support of this work by the UICC Research Board is appreciated, as is the cooperation of the UICC computer center in supplying computer time. The author wishes to thank Dr. R. D. Willett for communicating some of his results prior to publication. The cooperation of Drs. R. Caputo, S. Roberts, R. D. Willett, and B. C. Gerstein in agreeing to delay publication of their works so this paper could appear simultaneously is also appreciated. Registry No. [(CH3)3NH]3Mn2C17, 58150-41-7.

Holt et a).

References and Notes L. J. de Jongh and A. R. Miedema, AdL. Phys., 23, 1 (1974). See, e.g., S. Merchant, J. N. McElearney, G. E. Shankle, and R. L. Carlin, Physica (C’trecht),78, 308 (1974). R. M. Clay, P. Murray-Rust, and J. Murray-Rust, J . Chem. Soc., Dalton Trans., 595 (1973). R. Dingle, M. E. Lines, and S. L. Holt, Phys. Reo., 187, 643 (1969). L. R. Walker, R. E. Dietz, K. Andres, and S. Darack, Solid State Commun., 11, 593 (1972). M. T. Hutchings, G. Shirane, R. J. Birgeneau, and S. L . Holt, Phys. Rel;. B , 5 , 1999 (1972). R. E. Dietz, F. R. Merritt, R. Dingle, D. Hone, B. G . Silbernagel, and P. M. Richards, Phys. Rei;. Lett., 26, 1186 (1971). R. Caputo, S. Roberts, R. D. Willett, and B. C. Gerstein, Inorg. Chem., preceding paper in this issue. J . N. McElearney, D. B. Losee, S. Merchant, and R. L. Carlin, Phys. Rev. B, 7, 3314 (1973). T. Smith and S. A. Friedberg, Phys. Rev., 176, 660 (1968). J. W. Stout and W. B. Hadley, J . Chem. Phys., 40, 55 (1964). M. E. Fisher, Proc. R. Soc. London, Ser. A , 254, 66 ( I 960). J. N . McElearney, S. Merchant, G . E. Shankle, and R. L. Carlin, J . Chem. Phys., in press. Y. Tazuke and K. Nagata, J . Phys. Sot. Jpn., 38, 1003 (1975). H. W. White, K. H. Lee, J. Trainor, D. C. McCollum, and S. L. Holt, AIP Conf Proc., No. 18, 376 (1973). K. Takeda, Phys. Lett. A , 47, 335 (1974). R. E. Dietz, L. R. Walker, F. S. L. Hsu, W . H. Haemmerle, B. Vis, C. K. Chau, and H. Weinstock, Solid State Commtrn., 15, 1185 (1974). B. Vis, C. K. Chau, H. Weinstock, and R. E. Dietz, Solid State Commun., 15, 1765 (1974). K . Kopinga, T. de Neef, and W. J. M. de Jonge, Phys. Rev. 5 , 1 1 , 2 3 6 4 (1975).

Contribution from the Department of Chemistry, The University of Wyoming, Laramie, Wyoming 82071

Physical Properties of Linear-Chain Systems. 111. Absorption Spectra of RbFeBr3, CsFeBr3, RbFeC13, CsFeC13, and C~Mgi-~Fe~C13 C H A R L E S F. PUTNIK, G. M A T T N E Y COLE, Jr., B. B. GARRETT,I and S M I T H L. HOLT*

Received October 7 , 1975

AlC50734E The polarized, single-crystal absorption spectra of the linear-chain compounds RbFeBr3, CsFeBr3, RbFeC13, CsFeCI3, and C s M g 1 - ~ F e ~ Chave l 3 been measured between 298 and 4.2 K. The spectra of the pure salts and the dilute samples are similar in terms of absorption positions but the spin-forbidden transitions in the pure salts show anomalous intensity. The intensity of spin-forbidden absorptions in some cases nearly equals that of the spin-allowed sT2(D) 5E(D) absorption. The spectra are assigned to the octahedral, ligand field states of the high-spin d6 electronic configuration. The intensity of the spin-forbidden transitions results from a cooperative mechanism which relaxes the spin selection rule without energetically perturbing the electronic states involved. The results of this work add significant feature$ to the empirical characterization of this mechanism and may aid theoretical progress in this area

-

Introduction The unusual physical properties displayed by certain materials of the ABX3 formulation derive from the unique structure of these compounds. When A+ = Cs+, Rb+, or (CH3)4N+, B = a divalent, first-row transition metal ion, and X- = C1-, Br-, or I-, these materials crystallize as a hexagonal array of infinite linear chains of face-sharing [X3BX3]4octahedra separated by the A cations. The interchain separation afforded by the A cations acts as an effective magnetic insulation which constrains exchange interactions to one dimension, that along the chains.2 The unique character of these systems is evidenced in their magnetic and spectral properties. Tetramethylammonium trichloromanganate( 11), TMMC, orders in three dimensions only a t 0.84 K3 while spin correlation in one dimension is evident a t much higher temperatures.4 Measured absorption spectra reveal a cooperative mechanism which relaxes the hs = 0 spin selection rule and results in marked enhancement of absorptions corresponding to formally spin-forbidden transitions.3J-7 Such enhancement is observed in materials which possess isotropic exchange coupling but appears to be most pronounced in the ABX3 linear-chain materials.

In an effort to characterize these unique compounds more fully and to resolve better the effect of unidimensional magnetic exchange on the electronic properties of the system, the absorption spectra of RbFeBr3, CsFeBr3, RbFeC13, CsFeCb, and C ~ M g ~ F e i - ~ C have 1 3 been investigated between room temperature and 4.2 K. CsFeCl38 and RbFeC138-10 are reported to possess intrachain ferromagnetic coupling and interchain antiferromagnetic coupling with TN < 1.3 K and TN = 2.54 K, respectively. RbFeBrs is reported to possess intrachain and interchain antiferromagnetic coupling with 7” = 5 . 5 K. The magnetic properties of CsFeBr3 have not been studied in detail but measurements of powder susceptibility between room temperature and 4.2 K correlate well with those of RbFeBr3.12 Thus, these materials provide a means of comparing the effects of ferromagnetic and antiferromagnetic intrachain exchange coupling on the absorption spectra of isoelectronic, isostructural materials. This study also serves a second fundamental purpose by providing a better empirical characterization of the triplet structure of iron(I1). The electronic structure of this ion in an octahedral field is dominated by spin-forbidden 5T2(D) 3I’ transitions with only one quintet-quintet excitation possible.

-

Inorganic Chemistry, Vol. 15, No. 4, 1976 827

Linear-Chain Systems

Because the excited quintet state is subject to a dynamic Jahn-Teller distortion, the spin-allowed, 5T2(D) 5E(D), transition has received the majority of attention.13-1s Other -+

investigators have examined side bands in the FeF2 spectruml9-21 and only Ferguson et al. in their study of R b F e F P have treated the triplet levels in any detail. However, in this latter work measurements were carried out at 77 K where significant overlap complicates the spectral assignment. The strong enhancement of the triplet absorptions in the spectra of the AFeX3 materials facilitates assignment of these states and comparison of the spectra provides a useful empirical characterization of the mechanism of enhancement.

Experimental Section Preparation. Anhydrous FeCh was prepared by heating Mallinckrodt AR grade FeC1~4H20at 250 OC in a stream of dry HCI gas. Anhydrous FeBn (Great Western Inorganics), CsCI, RbCI, CsBr, RbBr (ROCJRIC), and MgCh (Ventron) were purchased at 99+% purity and all compounds were purified further by zone refining. The compounds CsFeCI3, RbFeCb, CsFeBr3, and RbFeBr3 were prepared by mixing equimolar quantities of the appropriate alkali halide and iron(I1) halide, sealing the mixture in an evacuated quartz ampule, and refining via a modified Bridgman method. CsMgCb was prepared in the analogous manner and combined with the appropriate molar quantity of CsFeC13 to produce the compounds CsMgo gsFeo.osC13 and CsMgo ssFeo.izC13. (Exact concentrations were determined spectroscopically.) In this manner, single-crystal boules - 5 cm in length and 1.5 cm in diameter were produced. From these it was possible to cleave sections suitable for u, r,and axial spectroscopic measurements. The crystals cleaved easily in planes parallel to the c axis but cleavage perpendicular to the c axis was difficult and the axial samples required careful polishing with a 9 5 5 methanol-water solution. When viewed with unpolarized light, crystalline samples of CsFeCh, RbFeCh, CsFeBr3, and RbFeBr3 appeared deep brown-amber. However, when observed with polarized light, the compoundsappeared deep red when the plane of polarization was parallel to the c axis and pale yellow when the plane of polarization was perpendicular to the c axis. The magnesium compounds containing small percentages of iron(I1) appeared faint brown-amber in unpolarized light, pale pink when viewed with plane-polarizedlight parallel to the c axis, and very pale yellow when the plane of polarization was perpendicular to the c axis. Crystal Structures. CsFeC13,23 RbFeCb,23 RbFeBrQ4.25 and CsMgCP are isomorphous and belong to the hexagonal space group P63Jmmc. The cell dimensions are a = 7.24, 7.06, 7.38, and 7.27 A and c = 6.05, 6.02, 6.28, and 6.19 A, respectively. The structure consists of linear chains of face-sharing [FeX6I4- or [MgC16I4octahedra separated by the alkali cations. The transition metal ion site symmetry is D3d. The structure of CsFeBr3 has not been reported; however preliminary x-ray powder diffraction measurements indicate this compound possesses the same structure.27 Spectroscopic Measurements. The u and T,single-crystal absorption spectra of CsFeCI3, CsFeBa, RbFeC13, and RbFeBn were measured at 298, 150, 77,40, 20, and 4.2 K in the spectral range 4500-30000 cm-1. The axial spectra were measured over the same range and were found to be identicalwith the u measurements indicating all transitions are of electric dipole nature. The u and s spectral measurementsof CsMgi-,FexCI3 were made at 298, 77, and 4.2 K in the range 4500-34000 cm-1. All spectra were measured using a Cary Model 14-RI recording spectrophotometer. The single crystals were mounted on aluminum rings which were inserted into the sample chamber of an Oxford Cf-100 cryostat. Temperature was determined and controlled to within A2 K with an Oxford VC-30 temperature controller. Polarized spectra were obtained by inserting GlanThompson prisms in the light paths of the spectrometer's sample and reference compartments. Oscillator strengths were determined as previously described?* The error in this method is estimated to be less than 10%. Results In terms of general appearance and absorption positions the spectra of RbFeBr3, CsFeBr3, RbFeC13, CsFeC13, and CsMgo.ssFeo.12C13 are similar and are shown in Figures 1-5, respectively. The spectrum of CsMgo.ssFeoosC13 is also shown

Table I RbFeC1, obsd energies, cm-l EIIc

Eic

CsFeC1, obsd energies, cm-' EIlc

Elc

Calcd

energies, Assignment

crn-l0

6400 6.211 7350 7350 7250 12658 1 2 3 4 6 11 993 14 599 15 152 1 5 1 0 6 15207 15378 15 444 1 5 686 16 671 -18200 18182 -18300 18250 19 128 19 186 19354 19357 19410 -19250 19649 20182 20157 20206 20161 20167 21 142 21 133 21 169 21 142 21 342 21 872 21 805 21 810 21 834 -22 625 -22 700 22 252 23 213 23 242 23 288 23 250 23419 23359 23 379 23425 23 923 2 3 4 6 9 23463 23474 23 940 23 923 23981 24115 24114 24190 24 155 26 116 6803 7519 12346

a Dq

6850 7576 12157

= 725 c m - ' ; B = 875 c m - ' ; C = 3 8 0 0 c m - l .

Table I1 RbFeBr, obsd energies, cm-'

CsFeBr, obsd energies, cm-I

EIlc

Eic

EIlc

Elc

5 848

5 882 6 711 12048 14 815

5 525 6589 12422 15 267

5 495 6661 12 195 15 152

17 699

17 191 17 687

6711 12048 14699 17277 17575 18 437 1 8 584 18702 18 847 19026 19209 19 342 19 685 20387 20781 21 368

22799 22899

17153 17730 18 116 18691 19004 19157 19 654 20445 20759 21 119 22 583 22768 22999 23 148 23 397

18 471 18 660 18 737 18 888 19 060 19 305 19 433 19 697 20243 20479 20768 21 137 21 400 21 950

Calcd

energies, Assigncm-Ia ment 6 500 11 800

'E(D) 3T,(H)a

16 357

'T2(H)

18931

3T,(P)

19 333

3T,(H)b

20789

,T2(F)

21 453

,E(H)

25 177

'T,(G)

18 748 19 073 19 220 19 673 20 243 20492 20756 21 160 22 22 22 23 23

609 805 946 191 419

aDq=650cm";B=900cm~';C=3600cm~'.

-

in Figure 5 but is of such low intensity above 10000 cm-1 that determination of absorption positions is difficult and ambiguous. The observed energy levels of the other systems are listed in Tables 1-111. The spectra are characterized by five absorption regions. The region from 4500 to 10000 cm-1 contains a broad, intense, asymmetric band attributable in all of the compounds to the spin-allowed 5T2 5E transition. It is the most intense region in the spectra of the dilute samples. This band decreases in oscillator strength by between 10% and 3096, depending on the compound, as the temperature is lowered from 298 to 77 K. Below 77 K the oscillator strengths remain constant within experimental error. This behavior parallels that of a vibronically assisted transition. The os-

-

828 Inorganic Chemistry, Vol. 15, No. 4, 1976

Holt et al.

Table 111 CsMg,. 86F%. IZC4 obsd energies, cm-'

Calcd energies, cm-] a

E II c

Elc

6 810 7 690 1 2 500 1 5 380

6 850 7 690 1 2 151 1 5 385

17 990 1 8 700 19 194 19 342 1 9 570 1 9 760

1 8 018

Assignment

7 250 11 995 16 801

20 408 20 534 20 661 21 186 21 277 21 505 22 727 23 256 23 390 23 4 6 0 23 580 23 644 23 714 24 155 24 390 25 400

1 9 342 1 9 569 19 755 20 202 20 421 20 534 20 665 21 186 21 277 22 4 8 0 23 256 23 364 23 419 23 458 23 568 23 663

20 280

21 530 22 410 IIbPeBrg 4.P.K

24 155 24 390 25 400 27 500 28 600 30 520

26 331 2 1 116 27 399 28 048 30 550

= 725 c m - ' ; B = 900 cm-'; C = 3800 cm-'.

cillator strength of this absorption in the materials studied is given in Table IV. Also presented in Table IV are the 3r manifolds demeasured oscillator strengths of the 5T scribed in the remainder of this section. The second major region of absorption lies between 10000 and 13000 cm-1. It contains a broad, weak manifold in both polarizations. The oscillator strength of the E I c component is about twice that of the E 11 c component in all of the spectra. The third region extends from 13000 to 17000 cm-I. It contains a broad, asymmetric, fairly intense manifold. The oscillator strength of the E 11 c absorption is approximately twice that of the E Ic absorption. In CsFeC13 this manifold resolves at 4.2 K into a low-energy shoulder and three peaks

-

-

-

W

L6

10

0

am4 x tal

Figure 1. Polarized, single-crystal absorption spectrum of RbFe. Br, at 4.2 K (optical density in arbitrary units),

on the band top. The spacing between the three peaks is -240 cm-1. These may be components of a vibrational progression corresponding to the 245-cm-1, Elu, infrared-active mode reported for C ~ F e c 1 3 .This ~ ~ is the only progression observed in any of the AFeX3 spectra. The higher energy regions of these spectra are well resolved only at 4.2 K. At room temperature the region above 17000

Table IV. AFeX, Oscillator Strengths ( f X l o 5 ) at 4.2 K Assigned components

Manifold center, cm-' -7 000

CsFeBr,

RbFeBr,

CsFeC1,

ElC

3.64 1.82 0.18 0.32

1.78 1.27 0.12 0.37

4.06 2.12 0.06 0.22

4.62 2.62 0.15 0.36

3.97 2.15 0.16 0.15

E 11 c El c

0.42 0.19

0.60 0.17

0.66 0.17

1.23 0.43

0.35 0.22

E /I c E l c

'TI ( W a

-12 300

;'::; }

-15000

:;g

-20 000

WH) 'T,(P)

E I1 c

E I1 ca

ELca

-9.5 -1

-1 1 -2

-6 -0.5

RbFeC1,

-8.5 -0.5

CsMg,. ,,Fe,. I $1,

-2.5 -1

WH)

-23 000

E /I c

-24 000

Elc E /I c E l c

'r(G)

3T2((3

0.49

0.89

0.65 0.17

0.15

0.19 0.11

0.27 0.26 0.08 0.16 a Overlap with adjacent d-d bands and the charge-transfer absorption negates the significant-figure accuracy of these measurements; however, relative values and magnitudes are significant.

Inorganic Chemistry, Vol. 15, No. 4 , 1976 829

Linear-Chain Systems

n

PO

10

16 cm" x

6

IO'

Figure 2. Polarized, single-crystal absorption spectrum of CsFeBr, at 4.2 K (optical density in arbitrary units).

cm-1 appears as a broad shoulder on the charge-transfer band. Partial resolution of some components in this region begins at 150 K. The fourth region lies between 17000 and 22500 cm-1. It contains a series of four intense, strongly overlapped bands in the pure systems and four much less intense absorptions in the dilute samples. In all cases in the pure systems the E 11 c absorptions are some 10-20 times more intense than the E I c counterparts. The E 1) c bands are - 5 times more intense in the dilute samples. The fifth region extends from -22500 cm-1 to the charge-transfer absorption edge. The E I c spectra contain two absorption manifolds in this region. The first manifold near 23000 cm-1 is resolved at 4.2 K into three sharp components in CsFeC13 and RbFeC13, two sharp components in RbFeBr3 and CsFeBr3, four sharp, weak components in CsMgo.ssFeo.izCh, and a single, extremely weak absorption in CsMgo.9sFeo.osC13. In all but the 5% sample these are the sharpest absorptions in the spectra. The manifold increases in oscillator strength by -50% as the temperature is lowered from 150 to 4.2 K in all spectra but that of the 5% sample. The majority of the increase occurs below 20 K. This absorption is fully resolved in the E 11 c orientation only in CsFeC13, RbFeBrs, and the dilute compounds. The second band in this region is observed near 24000 cm-1 in the E I c spectra. It appears as a rather weak absorption with a low-energy shoulder in CsFeCb and RbFeCls, as two sharp peaks in CsFeBrs and RbFeBr3, and as two very weak peaks in the 12% sample and is not resolved in the 5% sample. The intensity of this absorption shows a slight increase as the temperature is lowered from 77 to 4.2 K. The E 11 c component of this band is observed only in CsFeCb and the 12% sample. Overlap with the charge-transfer edge obscures this band in the other E I( c spectra. In the spectra of the pure compounds the charge-transfer absorption edge begins to dominate the spectrum at -24000

-

86

10

10

16

ri'x

I

n'

Figure 3. Polarized, single-crystal absorption spectrum of RbFeCl, at 4.2 K (optical density in arbitrary units).

cm-1 with E )I c and at -25000 cm-1 with E I c. In the dilute samples the E 11 c charge-transfer edge is observed at -25000 cm-1. In the E I c orientation the charge transfer in the dilute systems is less intense and much broader than the previous cases. In this orientation two additional absorptions are found near 25750 and 29500 cm-1. The oscillator strengths of the absorptions between 10000 and 22500 cm-1 in the pure samples appear to increase by -10% as the temperature is lowered from 150 to 77 K and then to decrease by the same amount as the temperature is further lowered to 4.2 K. These measurements were complicated by overlap between adjacent bands and overlap with the charge-transfer absorption. The estimations involved negate the absolute readings but the trend in oscillator strength as temperature is lowered is valid. The absorptions above 10000 cm-1 are dependent on concentration. The spectrum of the dilute sample containing 12% iron(I1) contains all of the absorptions present in the spectra of the pure systems. In the case of the 5% sample the limits of our apparatus allowed spectral measurement of a 5 mm thick crystal which was insufficient to resolve all absorptions. Finally, the axial spectrum of all compounds was measured and found to be identical with the E I c measurements. Assignment of the Spectra The major absorption regions of the AFeX3 spectra can be accounted for by considering the 5T2(D) 5E(D) transition and the numerous triplet, ligand field states of the d6 electronic configuration. The measured spectra were assigned by comparison with energies obtained from the computer diagonalization of the weak-field, octahedral, d6 matrices given

-

830 Inorganic Chemistry, Vol. 15, No. 4, 1976

Holt et al.

CsFeCI, 4.Z°K

20

25

16

5

10

cm-'x 10' Figure 4. Polarized, single-crystal absorption spectrum of CsFeC1, a t 4.2 K (optical density in arbitrary units).

..-., .'., *I

I'

'\\

CsMg,-,Fe,CI,

4.WK

L C

-..__-._

30

25

20

15

10

5

crn" x IO'

Figure 5 . Polarized, single-crystal absorption spectra of CsMg0.88Fe0.1 $1, and CsMgo.,,Fe0.,,C1,at 4.2 K (optical density in arbitrary units).

by Ferguson et a1.22 together with weak-field term energies given by Slater.30 These matrices include only the triplet states and do not include spin-orbit coupling. The crystal field strength, Dq,and Racah interelectron repulsion parameters,

B and C, were estimated from experimental results and input as variables. The calculated energies were then compared with observed energies until a best fit was obtained. The results are listed in Tables 1-111.

Inorganic Chemistry, Vol. 15, No. 4, I976 831

Linear-Chain Systems While this method proved useful in assigning some transitions, it clearly does not accurately account for all manifolds nor does it account for splittings and structure observed in the regions above -17000 cm-1. Konig and Kremer31 have published the results of the complete ligand field, Coulomb-repulsion, spin-orbit interaction calculation of the energy levels for the d6 electronic configuration in an octahedral field. This work indicates that the calculated state labeled 3T2(H) in our tables is 1000 cm-1 too high and that the state labeled 3T2(G) has components some 2000 cm-' lower in the complete calculation. These differences account for two major discrepancies between our calculation and measured spectra. The complete calculation also indicates that there is considerable splitting and state mixing in the region between -17500 and -21500 cm-1 which likely accounts for the additional peaks we have observed in this region. As Konig and Kremer did not publish the matrix elements with their diagram, no comparative calculation was made using the matrices which include spin-orbit coupling. We should also point out that the diagram furnished by Ferguson et al.22 with their matrices appears to be incorrect. Because of the manner of computation in that work, it was unclear from where the discrepancy arose. We have recalculated the matrices they presented and found them to be correct. Therefore, it appears that the three low-lying triplet states are incorrectly drawn in their diagram. The present labeling of 3T1(P), 3T2(H), 3T1(H) lowest to highest for the three low triplets should be revised to 3T1(H), 3T2(H), 3T1(P) to make the diagram useful. In addition to spin-orbit coupling, it is possible that some of the observed splitting may be due to the actual D3d site symmetry of the metal ion. In D3d the octahedral Ti state is split into A2 and E components while the T2 is split into an Ai and E. Likewise in the 0 3 double group the octahedral r4 is split into I'2 and r3, and I's is split into ri and r3. While this effect most likely makes some contribution to the observed splitting, the absence of clear-cut polarization behavior in the measured spectra makes unambiguous analysis on this basis impossible. A final consideration concerns the structured absorption manifold observed near 23000 cm-1 in all of the spectra and arises from transitions to spin-orbit components of 3G. This absorption manifold shows a large increase in oscillator strength as the temperature is lowered with the major portion of increase occurring between 20 and 4.2 K. The lack of total resolution above 20 K makes it impossible to follow individual components as temperature is varied. The behavior at least suggests the presence of magnon cold-band character in this absorption.

-

Discussion The most notable features of the spectra of the linear-chain AFeX3 compounds are the intensity of the formally spinforbidden transitions and the overall similarity of the spectra. The enhanced intensity makes available for study a large number of transitions which are not normally observed, while the similarities of the spectra aid in the spectral assignments and lend generality to the conclusions. The oscillator strength of the spin-allowed 5T2(D) 5E(D) transition is on the order of 10-5 in all compounds, as e x p t e d . The spin-forbidden transitions in the pure compounds possess oscillator strengths ranging from 10-6, for some transitions, to 10-5 for the transitions between 17000 and 22500 cm-1. For comparison, the oscillator strength of a spin-forbidden transition occurring id a metal ion in solution is on the order of 10-9. Therefore, the spin-forbidden transitions in the linear-chain AFeX3 compounds are enhanced by three to four orders of magnitude relative to those of isolated complexes and in some cases reach the intensity of spin-allowed transitions.

-

-

In the dilute system the spin-forbidden transitions are also enhanced but by one to three orders of magnitude less than the pure systems. The spin-forbidden transitions in the 12% sample are less intense by one order of magnitude and those in the 5% sample less by two to three orders of magnitude. Enhancement of this magnitude cannot be accounted for by single-complex, spin-orbit coupling and the temperature dependence of the oscillator strengths is not indicative of a vibronic interaction. It appears that a cooperative mechanism involving adjacent transition metal ions is dominant in these systems. Such a mechanism would account for the concentration dependence of the enhancement in that the dilute samples would contain fewer adjacent metal ions. Similar enhancement has been observed in certain lattices with three-dimensional coupling32 and a mechanism based on exchange-coupled pairs of metal ions has been postulated in explanation. The specifics of this mechanism have been discussed by several authors.lgJ2-43 The result of this mechanism, simply stated, is an exchange-coupled pair state from which magnon-exciton and double-exciton pair transitions can occur. Further, this mechanism results in a singleion, exchange-induced, electric dipole absorption process which is spin independent.35.38 While the relaxation of the spin selection rule by such a process is useful in qualitatively accounting for certain anomalies, it is not a complete explanation. An obvious point in fact is that transitions to states of the same symmetry and spin multiplicity differ in enhancement by as much as an order of magnitude. Indeed, as we discussed in a previous work,44 it is possible and at times necessary to involve two accepted mechanisms simultaneously to account for observed anomalous behavior. In order to elucidate the correct selection rules and more accurately to understand the interactions in pure salts which lead to such rules and enhancement, it would seem that a broader experimental characterization would be useful. Lohr and McClure34 initiated such a characterization based upon their study of a series of three-dimensionally coupled Mn(I1) salts. Many of their findings apply in general to the spectra of pure, magnetic salts. The results of this work allow us to expand on their characteristics and add several more. We shall restate the original characterization in general, rather than limit it to the manganese(I1) ion from which it derived, and add supportive or opposing evidence from our study. We will then include new characteristics derived from our work. The resulting empirical characterization of a cooperative interaction which is responsible for anomalous optical properties displayed by transition metal containing magnetic salts in the solid state can be summarized as follows. (1) Absorption intensity is relatively great whenever the metal ions are separated by a single ligand atom. When more than one ligand atom separates the metal ions, the observed intensity is closer to that of an isolated, single complex as would be the case for a polyatomic ligand like S042-. In the linear-chain systems where three monoatomic ligand bridges are present the pathway seems even more favorable as evidenced by the strong enhancement dominant along the chain axis. This is further supported by the decreased intensity observed in the dilute sample in which metal ion-metal ion interactions are significantly reduced. ( 2 ) The cooperative interaction mechanism breaks down spin-forbiddenness more effectively than single-complex spin-orbit coupling. The spin-orbit induced transition moment necessary to account for oscillator strengths of the order of 10-5 for spin-forbidden transitions would be absurdly large for d-d transitions in a centrosymmetric complex. ( 3 ) The intensity mechanism may destroy parity forbiddenness giving rise to sharp origin bands. This conclusion was based on a study of Mn(I1) spectra in which all transitions

832 Inorganic Chemistry, Vol. 15, No. 4, 1976 are spin and parity forbidden.34 In our work, however, direct comparison of spin-allowed and spin-forbidden transitions is possible and the spin-allowed intensity is not observably enhanced. If the parity selection rule were relaxed, enhancement of all d-d adsorptions to a level near that observed in tetrahedral complexes would be expected and this is not the case. (4) The cooperative mechanism does not give rise to temperature dependence of the type expected for vibronically assisted transitions described by f( T K ) = f ( 0 K ) coth (hvl 2kT) where v is the frequency of an odd-parity vibration. In the spectra reported here only the spin-allowed transition displays vibronically assisted temperature dependence. ( 5 ) The intensity enhancement is not related to the three-dimensional ordering temperature. The intensities of the pure AFeX3 crystals are anomalously large throughout the measured temperature range indicating that the principal cause of the enhancement is not related to long-range spin correlation. The lowest temperature at which our measurements were taken is 4.2 K and this is below the reported ordering temperature of only one of the studied compounds, RbFeBr3. There is no evidence of any spectral change in the 4.2 K spectrum of RbFeBr3, making it different from the other spectra. ( 6 ) The cooperative mechanism does not require translationally inequivalent metal ions. The large enhancements of the translationally equivalent iron complexes observed here certainly support this conclusion but no additional information on this point has been obtained. ( 7 ) The mechanism may intensify some crystal field bands more than others. The triplet absorptions between 17000 and 22500 cm-1 in AFeX3 spectra are enhanced an order of magnitude more than other triplet absorptions of the same state symmetry. To this initial characterization we can add the following points. (8) The cooperative mechanism does not appreciably alter the crystal field energy levels of the metal ion. The observed spectra, in terms of band positions, can be assigned according to rules derived for isolated, near-octahedral metal ion complexes. The absorption bands are intensified but not significantly shifted. (9) The cooperative mechanism is not dependent on the type of magnetic exchange dominant in the system. The spectra of intrachain ferromagnetic RbFeC13 and CsFeCls and intrachain antiferromagnetic RbFeBn and CsFeBn are different in no manner relatable to exchange. The observed differences resulted from the intrinsic difference in crystal field strength of chloride and bromide ligands. (10) The cooperative mechanism is dependent to some extent on the bonding pathway. The spin-forbidden transitions are enhanced more in the AFeBrs compounds than in the AFeC13 compounds. This may result from increased orbital availability, increased spin-orbit coupling, or increased polarizability in the bromides. (1 1) The Cooperative mechanism is not exclusively the property of pure compounds but is present when aggregates, possibly as small as a pair, exist. Enhancement is observed in the very dilute compound even though the number of linked ions is statistically small and is seen to increase as the metal ion concentration increases. This corresponds directly to an increasing number of linked metal ions as concentration is increased. Finally, the ease of observation of the enhanced triplet levels over a wide, spectral range makes these mea-

Holt et al. surements the best empirical characterization of the electronic structure of near-octahedral iron(I1) thus far obtained. This also suggests that improved characterization of spin-forbidden transitions in other metal ions may be obtained in systems where a cooperative enhancement mechanism is dominant. Acknowledgment. This work was supported in part by the National Science Foundation (Grant GP-47506) and the Office of Naval Research. Registry No. CsFeCb, 15611-69-5;RbFeC13, 15611-73-1; CsFeBr3, 53899-57-3; RbFeBr3, 37473-39-5; CsMgC13, 12371-11-8.

References and Notes ( I ) On leave from the Department of Chemistry, Florida State University,

Tallahassee, Fla. 32306. (2) For a review of experimental work on these systems see J. F. Ackerman, G. M. Cole, and S. L. Holt, Inorg. Chim. Acta, 8, 323 (1974), and references therein. (3) R. Dingle, M. E. Lines, and S. L. Holt, Phys. Reu., 187, 643 (1969). (4) (a) M. T. Hutchings, G. Shirane, R. J. Birgeneau, and S. L. Holt, Phys. Reu. B, 5, 1999 (1972); (b) R. J. Birgeneau, R. Dingle, M. T. Hutchings, G. Shirane, and S. L. Holt, Phys. Rev. Left., 26, 718 (1971). (5) J. F. Ackerman, E. M. Holt, and S. L. Holt, J . Solid State Chem., 9, 279 11974) (6) G. L: McPherson, T. J. Kistenmacher, J. 9 . Folkers, and G. D. Stucky. J . Chem. Phys., 57, 3771 (1972). (7) T. Li and G. D. Stucky, Acta Crystallogr., Sect. B, 29, 1529 (1973). (8) P. A. Montano, H. Schechter, E. Cohen, and J. Makovsky, Phys. Reu. B, 9, 1066 (1974). (9) P. A. Montano, E. Cohen, H. Shechter, and J. Makovsky, Phys. Reu. B, 7, 1180 (1973). ( I O ) M. Eibschutz, M. E. Lines, and R. C. Sherwood, Phys. Reu. B, 11,4595 (1975). M.E. Lines and M. Eibschutz, Phys. Recj. B, 11, 4583 (1975). S. L. Holt, to be submitted for publication. 0. G. Holmes and D. S. McClure, J . Chem. Phys., 26, 1686 (1957). F. A. Cotton and M. D. Meyers, J . A m . Chem. Soc., 82, 5023 (1960). W. E. Hatfield and T. S. Piper, Inorg. Chem., 3, 1295 (1964). G. D. Jones, Phys. Reu., 155, 259 (1967). G. Winter, Aust. J . Chem., 21, 2859 (1968). T. E. Freeman and G. D. Jones, Phys. Reu., 182, 41 1 (1969). Y. Tanabe and K.-I., Gondaira, J . Phys. Soc. Jpn., 22, 573 (1967). P. G. Russell, D. S. McClure, and J. W. Stout, Phys. Reu. Lett., 16, 176 (1966). T. Kambara, J. Phys. Soc. Jpn., 24, 1242 (1968). J. Ferguson, H. J. Guggenheim, and E. R. Krausz, Aust. J . Chern., 22, 1809 (1969). H.-J. Seifert and K. Klatyk, Z . Anorg. Allg. Chem., 324, 1 (1966). G. R. Davidson, M. Eibschutz. D. E. Cox. and V. J. Minkiewitz. AIP Conf. Proc;, No. 5 , 436 (1971). (25) M. Eibschutz, G. R. Davidson, and D. E. Cox, AIP Conf. Proc., No. 18, 386 (1973). (26) G. L. McPherson, T. J. Kistenmacher, and G. D. Stucky, J. Chem. Phys., 52, 815 (1970). (27) E. M. Holt, unpublished results. (28) C. Simo and S. L. Holt, J . Solid State Chem., 4, 76 (1972). (29) G. L. McPherson and J. R. Chang, Inorg. Chem., 12, 1196 (1973). (30) J. C. Slater, Quantum Theory At. Mol., Solid Stare, 2, Appendix 24 ( 1 966) --, (31) E. Konig and S. Kremer, J . Phys. Chem., 78, 56 (1974). (32) D. D. Sell, R. L. Green, and R. M. White, Phys. Rev., 158,489 (1967). (33) J. Ferguson, H. J. Guggenheim, and Y . Tanabe, Phys. Reu., 161, 207 \ - -

( 1 967)

(34) L.-L.’Lohr and D. S. McClure, J . Chem. Phys., 49, 3516 (1968). (35) J. Ferguson, H. J. Guggenheim, and Y .Tanabe, J . Appl. Phys., 36,1046 (1965). (36) J. Ferguson, H. J. Guggenheim, and Y . Tanabe, Phys. Reu. Lett., 14, 737 (1965). (37) Y. Tanabe, T. Moriya, and S. Sugano, Phys. Reu. Lett., 15, 1023 (1965). (38) J. Ferguson, H. J. Guggenheim, and Y. Tanabe, J . Phys. SOC.Jpn., 21, 692 (1966). (39) J. Ferguson, H. J. Guggenheim, and Y . Tanabe, J . Chem. Phys., 45, 1134 (1966). (40) K. Shinagawa and Y . Tanabe, J . Phys. SOC.Jpn., 30, 1280 (1971). (41) T. Fujiwara and Y . Tanabe, J . Phys. Soc. Jpn., 32, 912 (1972). (42) T. Fujiwara, W. Gebhardt, K. Pentanides, and Y . Tanabe, J . Phys. SOC. Jpn., 33, 39 (1972). (43) K. Ebara and Y . Tanabe, J . Phys. SOC.Jpn., 36, 93 (1974). (44) G. M. Cole, Jr., C. F. Putnik, and S. L. Holt, Inorg. Chem., 14,2219 (1975).